Learners need to perceive, comprehend, and translate movement information during skill acquisition. Several authors have suggested that point-light displays (PL) may be useful in conveying visual information to the learner. However, researchers who have compared the effectiveness of PL and normal video (VID) demonstrations have produced equivocal findings (Horn et al., 2001, 2003; Scully & Carnegie, 1998). Scully and Carnegie (1998) found that PL facilitated the acquisition of a gymnastic movement, whereas Horn et al. (2001, 2003) reported no differences between VID and PL in the acquisition of a soccer-chip task. The contradictory results may arise because the soccer-chip is a simple, single limb task, whereas the gymnastic task is a complex, multi-limb task. The benefits of PL over VID may only be apparent with tasks that involve multiple components. In the present study, the relative effectiveness of PL and VID demonstrations was compared using a complex, multi-limb-task. The degree of similarity in movement form between the model and the learner was taken as a measure of effectiveness. Sixteen adult males were assigned to a PL or VID group. Participants watched and imitated an under-arm bowling action 10 times. Quality of imitation was assessed using separate Likert-type rating scales for whole body movement and individual components of the action. The quality of imitation improved for both groups across acquisition, p < 0.01. Mean imitation scores showed that both the VID (M = 14.40, SD = 1.2) and PL (M = 12.16, SD = 3.12) participants produced similar movements to the model (max score = 16). The group effect approached significance, p = 0.056. Individual component analysis revealed a significant difference on the lunge component (initial step), p < 0.01. The PL group executed an incorrect left leg lunge compared with the VID group. The PL demonstration offered no advantages over VID when attempting to imitate a complex, multi-limb-task, although both groups produced movement patterns that were similar to the criterion model. The removal of structural information in PL may make it harder to perceive depth compared with VID. Implications for practice and instruction are highlighted. Keyword(s): coaching, performance, research